Von Duprin LLC v. Moran Electric Service, Inc.

In the United States Court of Appeals For the Seventh Circuit ____________________ Nos. 20-1711 & 20-1793 VON DUPRIN LLC, Plaintiff-Appellee, Cross-Appellant, v. MAJOR HOLDINGS, LLC and MAJOR TOOL AND MACHINE, INC., Defendants-Appellants, Cross-Appellees, and MORAN ELECTRIC SERVICE, INC., Defendant, Cross-Appellee. ____________________ Appeals from the United States District Court for the Southern District of Indiana, Indianapolis Division. No. 1:16-cv-01942 — Tanya Walton Pratt, Chief Judge. ____________________ ARGUED MARCH 30, 2021 — DECIDED SEPTEMBER 3, 2021 ____________________ Before KANNE, BRENNAN, and SCUDDER, Circuit Judges. SCUDDER, Circuit Judge. This is a complex environmental cleanup case out of Indianapolis, Indiana. Over several 2 Nos. 20-1711 & 20-1793 decades, owners and operators of industrial facilities dis- posed of chemical solvents used in manufacturing processes. The solvents have degraded over time and have seeped into the groundwater and soil in the surrounding residential area. Investigations showed that vapors emitting from the under- ground contamination have intruded into homes and a local park. Von Duprin LLC, whose predecessor in interest manufac- tured security hardware and once owned property in the area, undertook cleanup efforts and then sought to recover some of those costs as well as future remediation costs. Von Duprin sued former and current owners and operators of adjoining properties under the Comprehensive Environmental Re- sponse, Compensation, and Liability Act—a federal environ- mental statute often shorthanded as CERCLA. Following a bench trial, the district court found that Von Duprin and two other former or current owners and operators in the area bore responsibility for portions of the environmental harm. The court then assigned liability among and between all three par- ties. Before us now are an appeal and cross-appeal from lengthy proceedings in the district court. While we see no error in many of the district court’s rul- ings, we vacate the court’s threshold determination under § 107(a) of CERCLA that liability for remediating the environ- mental harm is divisible—capable of being apportioned on the basis of principles of causation—among and between the parties to this litigation. In the end, then, we affirm in part and vacate and remand in part for additional proceedings. Nos. 20-1711 & 20-1793 3 I A. The Properties at Issue The environmental harm developed over at least the last 50 years. The four relevant and adjacent properties, each in northeast Indianapolis, have changed hands over the years. We start with the Von Duprin Property. Von Duprin LLC’s predecessor, Von Duprin Inc., owned the property for over 40 years and during that time manufactured security hardware and safety products. The firm used degreasers and a variety of chemical products, including trichloroethylene (TCE) and perchloroethylene (PCE). Operations ceased in 1986, and a manufacturer named Threaded Rod Company, Inc. acquired the Von Duprin Property in 1987. Next up is the Moran Property. Moran Electric Service Inc., a company that used degreasers as well as TCE and other cleaning agents in connection with repairing electrical mo- tors, owned and operated this property from 1946 to 1996. In 2005, a company called Major acquired the Moran Property. Now the Ertel Property. Ertel Manufacturing, a manufac- turer of automotive engine parts, owned this property from 1917 to 1998. The firm used chlorinated solvents like TCE and PCE in its manufacturing processes. The Ertel Property changed hands in 1998, only later to be abandoned. In 2007 Major leased the Ertel Property from the City of Indianapolis. Major acquired full ownership in 2013. The Zimmer Paper Property, itself subdivided into the Zimmer Paper Facility and the Zimmer Packaging Facility, is the fourth property. Zimmer Paper Products, Inc. owned the property from 1986 to 2006, during which time the company used and disposed of chemical solvents. These properties 4 Nos. 20-1711 & 20-1793 were owned and operated by Moran from 1967 to 1984. Major acquired the Zimmer Paper Facility in 2007 and later acquired the Zimmer Packaging Facility in 2013. Major, which we use for simplicity in referring to Major Holdings LLC and Major Tool & Machine, is the current owner of the Ertel Property, Moran Property, and Zimmer Pa- per Property. No one alleges that Major released any hazard- ous materials at any of its properties, but current owners or operators of a site where hazardous materials were released may be held liable under CERCLA without having caused a release. So the overall ownership history breaks down this way: Properties Former Owner(s) Current Owner Von Duprin Property Von Duprin LLC Threaded Rod Moran Property Moran Electric Major Service Ertel Property Ertel Manufacturing Major Zimmer Paper Moran Electric Major Property (divided into Service the Zimmer Paper Zimmer Paper Facility and the Products Zimmer Packaging Facility) All agree that, at various times over the twentieth century, chlorinated solvents, including PCE and TCE, were dumped at and around the Von Duprin, Moran, Ertel, and Zimmer Properties. These solvents are toxic and over time they seeped Nos. 20-1711 & 20-1793 5 into the soil and groundwater, eventually commingling into a groundwater plume—a collection of groundwater mixed with hazardous contaminants underground—which flows from northeast (where the properties are located) to south- west, covering approximately three-quarters of a mile. Of par- ticular concern is that vapor emissions from the plume can migrate upwards from the groundwater and through the soil, resulting in vapor intrusion into structures. About 40 homes and an indoor facility at the public park have experienced this intrusion. B. Von Duprin’s Lawsuit After several environmental assessments and investiga- tions of the Von Duprin Property, the Indiana Department of Environmental Management or IDEM, determined in March 2009 that soil and groundwater at and around the property had been contaminated by chlorinated solvents. In 2013 IDEM informed Von Duprin that it could be a potentially responsi- ble party or PRP under Indiana law. Von Duprin then began its own investigation into the contamination. In time the com- pany entered Indiana’s cleanup program and performed sub- stantial remediation in the area. After it, too, received a poten- tial liability notice from IDEM and began to take remedial ac- tion, Threaded Rod, the current owner of the Von Duprin Property, filed a suit against Von Duprin and others to re- cover costs. That case later settled. Von Duprin has continued to undertake remediation ef- forts at a cost of $3.2 million. In 2016 the company invoked § 107(a) of CERCLA and turned to federal court to recover those costs from current and former owners of adjacent prop- erties—Moran, Zimmer, and Major. 6 Nos. 20-1711 & 20-1793 Moran and Major responded by going on the offensive and filing counterclaims and crossclaims against Von Duprin under § 113(f) of CERCLA. Zimmer never appeared or re- plied to the original complaint, so the district court entered a default against the company under Federal Rule of Civil Pro- cedure 55(a). C. An Overview of CERCLA Some background on CERCLA is essential to understand- ing the district court’s decision and issues before us on appeal. Over many decades, Congress has created legal mecha- nisms to encourage cleanup and continued stewardship of real property. A prime example came with the enactment of CERCLA in 1980. Congress designed the statute to “promote the timely cleanup of hazardous waste sites and to ensure that the costs of such cleanup efforts were borne by those respon- sible for the contamination.” Burlington Northern & Santa Fe Ry. Co. v. United States, 556 U.S. 599, 602 (2009) (cleaned up). The statute operates in part by encouraging, and directly providing the vehicle for, private parties to invest in environ- mental response, including remediation, and then to recover at least part of those response costs from other potentially re- sponsible parties or PRPs. In United States v. Atlantic Research Corporation, the Su- preme Court held that § 107(a) of CERCLA, codified at 42 U.S.C. § 9607(a), provides PRPs who incurred response costs with a cause of action to recover certain of those costs from other PRPs. See 551 U.S. 128, 131, 139 (2007). To establish a claim for cost recovery under § 107(a), a plaintiff must show that (1) the site in question is a “facility”; (2) the defendant qualifies as a PRP; (3) the facility experienced a release or Nos. 20-1711 & 20-1793 7 threatened release of hazardous substances; and (4) the plain- tiff incurred costs consistent with the National Contingency Plan in responding to the release. See Env’t Transp. Sys., Inc., v. ENSCO, Inc., 969 F.2d 503, 506 (7th Cir. 1992). Congress defined PRPs to include four entities: current owners and operators of a site that experienced a disposal of hazardous material, past owners or operators at the time of the release, persons who arranged for disposal of a hazardous substance at a site, and parties who transported a hazardous substance to a site. See 42 U.S.C. § 9607(a)(1)–(4). Everyone agrees that each party to this litigation qualifies as a PRP un- der CERCLA. Joint and several liability is the norm for PRPs under § 107(a). The exception is divisible liability—commonly called apportioned liability—where liability is assigned to PRPs ac- cording to the portion of the underlying environmental harm each caused. See Burlington Northern, 556 U.S. at 613. PRPs who find themselves sued under § 107(a) often file a counterclaim against the original plaintiff on the basis that the party is itself a PRP who caused part of the harm and thus should contribute to any ultimate remediation liability. These so-called contribution counterclaims proceed under § 113(f) of CERCLA, which authorizes the allocation of liability based on “such equitable factors as the court determines are appro- priate.” 42 U.S.C. § 9613(f)(1). At this point, what is essential to recognize is that appor- tioned (divisible) liability and allocated liability are not one and the same under CERCLA. To the contrary, they are dis- tinct, with apportioned liability imposed on the basis of 8 Nos. 20-1711 & 20-1793 principles of causation and allocated liability the product of an application of equitable factors. Two more background points help frame the issues on ap- peal. First, not all remediation costs are eligible for recovery under § 107(a). Rather, Congress has limited recovery costs to those incurred consistent with the National Contingency Plan or NCP. See 42 U.S.C. § 9607(a)(4)(B). The NCP establishes how private parties can best perform removal and remedial actions to ensure the achievement of CERCLA-quality cleanup. See 40 C.F.R. § 300. Second, CERCLA provides an affirmative defense to reme- diation liability to PRPs who did not release hazardous mate- rials themselves but remain potentially liable by virtue of be- ing a current owner of a site where a prior release occurred. Added in 2002, Congress called this defense the “bona fide prospective purchaser” or BFPP defense. See 42 U.S.C. § 9607(q)(1)(C). To qualify, a PRP must, in connection with the original purchase of the property in question, and among other things, make “all appropriate inquiries into the previ- ous ownership and uses of the facility in accordance with gen- erally accepted good commercial and customary standards and practices.” 42 U.S.C. § 9601(40). In that way, the BFPP de- fense shields from CERCLA liability those who invest in con- taminated lands and exercise diligence, do not impede cleanup efforts, and meet ongoing obligations. D. Summary Judgment After a period of initial discovery, the parties filed cross motions for summary judgment. The district court’s ruling on the divisibility of remediation liability—whether the norm of joint and several liability for PRPs should apply or instead Nos. 20-1711 & 20-1793 9 whether each PRP should bear responsibility to the degree they caused the underlying environmental harm—is a sub- stantial issue on appeal. In granting in part Moran and Ma- jor’s motions for summary judgment, the district court deter- mined that CERCLA liability was divisible and thus should be apportioned among and between the PRPs. The district court’s reasoning rested on two abbreviated observations. First, the court observed (without much accom- panying explanation) that the harm caused by the PRPs (Von Duprin, Major, Moran, and Zimmer) was “theoretically capa- ble of apportionment” and therefore divisible among them. From there, and second, the district court pointed to Moran’s expert witness, Dr. Adam Love, and, again without much elaboration, observed that there was a “reasonable basis for apportionment.” Who caused what harm and how much each PRP, including Von Duprin, should contribute to the overall remediation costs would, the district court concluded, be de- cided at a bench trial. At summary judgment the district court also concluded that Major was a BFPP and thus could not be held liable for response costs that Von Duprin (as the plaintiff) sought to re- cover relating to the Moran Property and Zimmer Packaging Facility. E. The District Court’s Bench Trial A six-day bench trial ensued in the summer of 2019. A bat- tle of competing experts—all offering different perspectives on which PRP caused what portion of the overall environ- mental harm—defined much of the trial. Other witnesses tes- tified about methods for allocating liability and whether the 10 Nos. 20-1711 & 20-1793 remediation costs incurred by Von Duprin were consistent with the National Contingency Plan. The district court’s ultimate Findings of Fact and Conclu- sions of Law included these key rulings: Recoverable Costs. The district court evaluated what portion of Von Duprin’s costs were incurred consistent with the NCP and therefore recoverable under § 107(a) of CERCLA. The court concluded that $1.7 million of the $3.2 million Von Duprin originally sought was recoverable, excluding the $1.5 million settlement with Threaded Rod from a previous law- suit and $39,000 for the costs related to IDEM oversight. Apportionment of Liability. The district court began by re- newing its prior observation (from summary judgment) that the environmental harm was “theoretically capable of appor- tionment.” It then found that, based on the evidence pre- sented at trial, a “reasonable basis for apportionment exists.” The court did so by crediting Moran’s expert Dr. Adam Love’s testimony as to how the harm could be apportioned among the properties—even while noting that the nature of the con- tamination made it impossible to differentiate the chemical compounds once they enter the groundwater plume and eventually vaporize. In a section of its Conclusions of Law labeled “Apportion- ment,” the district court held that Major was liable under § 107(a) as the current owner of the Ertel Property and Zim- mer Paper Facility. Von Duprin, the court further concluded, was liable as owner of the Von Duprin Property at the time of a hazardous release. Moran was liable too, because it owned the Moran Property at the time of a release. Since no party Nos. 20-1711 & 20-1793 11 offered evidence of a release when Zimmer owned the facil- ity, the district court assigned no liability to it. Allocation of Liability. In a section labelled “Allocation,” the district court then performed what it called “equitable alloca- tion” by applying a series of factors, including the distin- guishability of each party’s hazardous waste, the amount of that waste, the toxicity of the contaminants at issue, the de- gree of involvement of the parties in generating the waste, the extent of each party’s cooperation with environmental au- thorities to mitigate harm, and the overall care taken by each party in handling the hazardous waste. These factors are com- monly called the “Gore factors,” named after then-Congress- man Albert Gore who proposed them as an amendment to a 1980 Superfund bill. See Env’t Transp. Sys, 969 F.2d at 508. After listing these Gore factors, but without any explana- tion as to whether or how each factor played into its allocation analysis, the district court “apportioned” the response costs as follows: 50% to the Von Duprin Property, 10% to the Ertel Property, 20% to the Zimmer Paper Property, and 20% to the Moran Property. Each party was held responsible for 100% of the harm attributed to the property or properties that they once owned or now own. The district court then took those percentages and applied them to the total amount of Von Duprin’s recoverable costs—$1.7 million—and found that Von Duprin was responsible for $850,000, Major for $510,000, and Moran for $340,000. The district court considered the cross- and counterclaims filed by the parties to be resolved by these liability allocations. From there the district court ruled on Von Duprin’s re- quest for a declaratory judgment entitling it to recover antici- pated future remediation costs related to ongoing cleanup 12 Nos. 20-1711 & 20-1793 contamination at the adjoining properties. The court found an award of future costs warranted because the company had es- tablished its right to recover from the other PRPs for past re- sponse costs. The district court therefore awarded Von Duprin future costs in an amount to be determined based on the same allocation percentages of liability for past costs among the parties, conditioned on the company proving that subsequent costs were both incurred consistent with the NCP and related to contamination at the surrounding properties. Major’s BFPP Defense. The district court further considered Major’s request for a BFPP defense. At summary judgment, the district court had concluded that Major was a BFPP (and therefore shielded from liability under § 107(a)) for any harm arising from the Moran Property and Zimmer Packaging Fa- cility. Based upon the evidence presented at trial, however, the district court reached the opposite conclusion for Major’s two other properties, the Ertel Property and Zimmer Paper Facility. Major was liable, the district court continued, for harm resulting from those two properties because the com- pany could not show that it had satisfied the “all appropriate inquiries” requirement of the BFPP defense. Remaining State Law Claims. Finally, the district court sum- marily disposed of Von Duprin’s state law claims, observing that ruling in the company’s favor on those claims would amount to a prohibited double recovery. F. Major’s Appeal and Von Duprin’s Cross-Appeal Major appealed and Von Duprin filed its own cross-ap- peal. Major appeals the district court’s decision to deny it the benefit of the BFPP defense to liability for its ownership of the Nos. 20-1711 & 20-1793 13 Zimmer Paper Facility and the Ertel Property. Major also maintains that Von Duprin failed to prove its response costs were incurred consistent with the NCP. Finally, Major chal- lenges the district court’s calculation of damages, contending that the district court conflated the allocation and apportion- ment processes, resulting in a damages award unsupported by the evidence. Von Duprin cross-appealed, arguing that the district court’s liability assignment erred at an earlier step: it should never have concluded that the harm was capable of appor- tionment. In Von Duprin’s view, the district court should have held the defendants jointly and severally liable. At a more granular level, Von Duprin also challenges the district court’s failure to exclude Dr. Love’s testimony as unreliable. For its part, Moran asks us to affirm the district court in full. II We begin with a word on appellate jurisdiction. Upon docketing Major’s appeal and Von Duprin’s cross-appeal, we questioned whether the case had reached finality in the dis- trict court and directed the parties to brief the question. What concerned us was that the district court’s judgment made no mention of Zimmer’s default, how the court resolved Moran and Major’s counterclaims, and the amount of prejudgment interest owed Von Duprin as part of its damages award. In these ways, the district court’s judgment was deficient, as it invited the impression that the court had entered a partial judgment under Federal Rule of Civil Procedure 54(b) and thus not resolved all claims against all parties. See Gen. Ins. Co. of America v. Clark Mall Corp., 644 F.3d 375, 378–80 (7th Cir. 14 Nos. 20-1711 & 20-1793 2011) (explaining scenarios in which 54(b) judgments provide sufficient finality for purposes of appellate jurisdiction). Aided by the parties’ briefing of the finality question, we are confident in our jurisdiction. The district court’s final judgment order expressly accounted for 100% of the allocated liability by assigning amounts owed to the pertinent parties, thereby necessarily resolving Von Duprin’s primary claim under § 107(a) as well as Moran and Major’s counterclaims for contribution under § 113(f). We interpret the district court’s silence on Zimmer to mean—as everyone agrees—that Zim- mer, despite its default, owes nothing under the terms of the judgment. To be sure, the district court should have entered formal default judgment against Zimmer under Rule 55. And while the district court also should have specified the pre- judgment interest owed Von Duprin, the failure to do so does not indicate a lack of finality because the computation is “readily ascertainable from the record through only a minis- terial calculation” and, even more, the docket shows that Mo- ran has since paid the amount owed. Student Loan Marketing Ass’n v. Lipman, 45 F.3d 173, 177 (7th Cir. 1995). In these circumstances, then, we have no doubt the case reached finality in the district court and that we may proceed to the merits of the questions presented on appeal. III On the merits, we start with Von Duprin’s challenge to the district court’s threshold determination at summary judg- ment that remediation liability should not be joint and several but instead apportioned among Major, Moran, and Zimmer as the PRPs according to the environmental harm caused by each company. Major challenges the district court’s resolution Nos. 20-1711 & 20-1793 15 of its counterclaim seeking contribution from Von Duprin un- der § 113(f) of CERCLA for its own share of the remediation liability. Major sees the district court’s ultimate allocation of liability as unsupported by the evidence presented at the bench trial. Both Von Duprin and Major are right to find fault with the district court’s liability determinations. We see two errors. First, the district court applied incorrect substantive and pro- cedural standards in concluding at summary judgment that the environmental harm and attendant CERCLA liability could be apportioned. Second, after the bench trial, the district court conflated the concepts of apportionment and allocation and left unexplained the factors guiding its ultimate assign- ments of liability. These two errors interrelate in ways that leave us no choice but to vacate the district court’s judgment and to remand for further proceedings. A. Apportionment at Summary Judgment Section 107(a) of CERCLA imposes “strict liability” on PRPs for response costs resulting from a release of hazardous materials. Burlington Northern, 556 U.S. at 608. While the stat- utory text is silent on the scope of liability, the Supreme Court has explained that Congress intended that question to “‘be determined from traditional and evolving principles of com- mon law.’” Id. at 613 (quoting United States v. Chem-Dyne Corp., 572 F. Supp. 802, 808 (S.D. Ohio 1983)). The primary guideposts come from the Restatement (Sec- ond) of Torts. And we know from the Restatement that “where two or more persons cause a single and indivisible harm, each is subject to liability for the entire harm.” Chem- Dyne, 572 F. Supp. at 810 (citing Restatement (Second) of 16 Nos. 20-1711 & 20-1793 Torts, § 875). The imposition of joint and several liability holds each party responsible for the entirety of the harm. See id. The Restatement explains that liability may be divided— or apportioned—“when two or more persons acting inde- pendently caused a distinct or single harm for which there is a reasonable basis for division according to the contribution of each,” with each responsible party being “subject to liabil- ity only for the portion of the total harm that he has himself caused.” Id. (citing Restatement (Second) of Torts § 433A). Be- cause “[n]ot all harms are capable of apportionment,” the Su- preme Court has emphasized that “CERCLA defendants seeking to avoid joint and several liability bear the burden of proving that a reasonable basis for apportionment exists.” Burlington Northern, 556 U.S. at 614. The facts and evidence must permit a “reasonable basis” for dividing liability among PRPs, as “arbitrary apportionment for its own sake” is incon- sistent with CERCLA’s liability scheme. See id. at 614 (cita- tions omitted). As a practical matter, joint and several liability often re- flects the norm in complex environmental cleanup cases be- cause most circumstances reveal numerous disposers of waste over long periods of time where after-the-fact identification of who contributed what and thus who caused what portion of the present-day harm at issue is exceptionally difficult to as- certain with reliability. A leading case in this area—Chem- Dyne, from which the Supreme Court took direction in Bur- lington Northern—reinforced this precise reality. See Chem- Dyne, 572 F. Supp. at 810–11 (declining at summary judgment to determine that liability could be apportioned because the evidence, when viewed in the light most favorable to the non- Nos. 20-1711 & 20-1793 17 moving party, showed that multiple parties had contributed to the contamination and over time much of the waste had become commingled, making the divisibility determination all the more complex). Our cases, too, have recognized that joint and several liability is the norm in complex CERCLA cases. See Metro. Water Reclamation Dist. v. N. Am. Galvanizing & Coatings, Inc., 473 F.3d 824, 827 n.3 (7th Cir. 2007) (“The only exception to joint liability is when the harm is divisible, but this is a rare scenario.”); United States v. Capital Tax Corp., 545 F.3d 525, 535 (7th Cir. 2008) (“Divisibility is the exception, however, not the rule.”). The apportionment inquiry proceeds in two steps. See Burlington Northern, 556 U.S. at 614 (laying out the two-step process); United States v. NCR Corp., 688 F.3d 833, 838 (7th Cir. 2012) (applying that process) (“2012 NCR Corp.”). The thresh- old question is whether the harm is even capable of appor- tionment. See Burlington Northern, 556 U.S. at 614. Making that determination requires an initial look at the underlying facts to determine “what type of pollution is at issue, who contributed to that pollution, how the pollutant presents itself in the environment after discharge, and similar questions.” 2012 NCR Corp., 688 F.3d at 838. If the preliminary review of the facts suggests that the harm can be apportioned, the in- quiry then must take the second step of asking with even more factual particularity whether the evidentiary record provides a reasonable basis for actually apportioning the lia- bility among the PRPs. See id. Although the case law has described these inquiries as proceeding in two distinct steps, it is often easier and more practical to think of them as blending together and interre- lated. Whether undertaken in one or two steps, the 18 Nos. 20-1711 & 20-1793 overarching question is whether, as a matter of record evi- dence, the environmental harm (and responsibility for the re- sponse costs at issue) can be apportioned by applying princi- ples of causation. Whether the evidence permits the apportionment of reme- diation liability is itself often a contested question. Some par- ties may insist the answer is yes, while others will urge the district court to say no. That is what happened here: Major and Moran contended that the evidence in the summary judg- ment record allowed the district court to make findings as to which PRP caused what environmental harm. Von Duprin took the opposite position, arguing that the same record evi- dence provided no reliable basis for making the findings nec- essary to apportion the liability. The district court’s responsi- bility at summary judgment was to ensure that Major and Mo- ran, as both the moving parties and the parties with the bur- den of showing that cost-recovery liability is divisible, had identified materially uncontested record evidence allowing that question to be resolved in the affirmative without a trial. See Burlington Northern, 556 U.S. at 614 (explaining that the burden for showing divisibility is on the moving party); Fed. R. Civ. P. 56(a) (identifying the burden on the party moving for summary judgment). The point of this observation is to recognize and rein- force—as the district court did in Chem-Dyne—the complexity and intersection of the substantive liability question with the procedural posture at which the question is being asked. The apportionment inquiry must be answered at summary judg- ment in a manner consistent with both the standards (and pro- tections) of Rule 56 and the substantive burden imposed by § 107(a) of CERCLA on the party requesting that cost- Nos. 20-1711 & 20-1793 19 recovery liability be divided among the PRPs instead of joint and several. The district court did not approach the apportionment question this way at summary judgment. In granting Moran’s motion for partial summary judgment on the apportionment question, the district court determined that the harm was “theoretically capable of apportionment” based on the opin- ion offered by Moran’s expert, Dr. Adam Love. He opined that there were distinct sites of pollution at each property that could be used to effectively account for how much of the harm emanated from each property. This was possible, Dr. Love ex- plained, by examining the chemical composition of soil sam- ples from those source sites to ascertain the concentration of certain chemicals at each site and to compare those respective concentrations to the chemical composition of the groundwa- ter plume. In Moran’s view, Dr. Love’s analysis established that the Von Duprin Property’s contributions to the total harm were distinct from that of the other properties. This was so, according to Dr. Love, because soil samples from the Von Duprin Property showed a larger presence of PCE, whereas the soil samples from the other properties—which sit upgra- dient from Von Duprin, meaning that the groundwater flows down toward the Von Duprin Property—show a larger pres- ence of TCE. Thus, in Dr. Love’s opinion, responsibility for the environmental harm at issue was reasonably apportiona- ble based on the chemical composition at the source sites vis- à-vis the composition of the plume. Von Duprin disagreed, maintaining that liability should be joint and several because the complexity of the underlying facts showed that the harm was incapable of reliable appor- tionment. The company supported this view by attacking Dr. 20 Nos. 20-1711 & 20-1793 Love’s opinion and raising larger questions about the feasibil- ity of dividing the harm given the complexity of the underly- ing facts. Dr. Love’s methodology, Von Duprin contended, re- lied on soil samples collected well after large quantities of soil had been removed from the properties and decades after re- leases of hazardous waste. Von Duprin further offered its own expert, John McInnes, who opined that the four properties all played a part in releasing chlorinated solvents over many dec- ades. Von Duprin also urged that apportionment was inap- propriate because the toxic chlorinated solvents became indis- tinguishable once they commingled in the groundwater plume. These observations and facts, Von Duprin advanced, precluded a finding of divisibility at summary judgment. The district court reacted by seeing Von Duprin’s position as a “concern [that] highlights a practical problem, not a the- oretical one.” Indeed, the district court went even further, ob- serving that Von Duprin’s concerns with the reliability of Dr. Love’s apportionment analysis are “irrelevant to the question of whether harm is divisible.” Therein lies the misstep. The district court’s approach re- flects two errors—one substantive and one procedural—that interrelated in a way that sent the incorrect question to trial and had the effect of relieving Major of its burden of showing that liability should be apportioned instead of imposed jointly and severally on each PRP. The errors were the product of the district court approach- ing the divisibility question at too high a level of generality, taking too literally the use of the word “theoretically” in our decision in 2012 NCR Corp. See 688 F.3d at 838. There we re- stated the Burlington Northern test in terms of two steps. We described the first step as one at which “we must determine Nos. 20-1711 & 20-1793 21 whether the harm at issue is theoretically ‘capable of appor- tionment.’” Id. (quoting Burlington Northern, 556 U.S. at 614). The district court overread the adverb’s importance to the test. After all, nearly everything in theory is capable of divi- sion if examined at a high enough level of generality. But see- ing the question in theoretical—rather than factual and evi- dentiary—terms robs the necessary apportionment inquiry of any meaningful content and leads to an analysis divorced from the Supreme Court’s instructions in Burlington Northern and the guiding principles in the Restatement (Second) of Torts. Indeed, we made much the same observation in 2012 NCR Corp. itself. We emphasized that whether the environmental harm at issue was capable of apportionment depended—at the first step of the analysis—on many findings of fact. See 688 F.3d at 838. Here, however, the district court seemed to con- clude that the remediation liability was divisible—at least at the theoretical level—because Moran’s expert Dr. Love of- fered that view. But stopping with that observation effectively transformed the norm of joint and several liability in CERCLA cost-recovery cases to divisible liability. Put another way, our review of the summary judgment record leaves us of the view that the district court stopped short of grappling with the complexity inherent in a CERCLA cost-recovery case as wide- ranging as this one. We do not read the Supreme Court’s Bur- lington Northern decision as intending every asserted appor- tionment defense to so easily clear the evidentiary hurdle de- manded by the standard embodied in § 433A of the Restate- ment. Consider the complexity of the environmental harm at is- sue. At least three different owners of four different properties 22 Nos. 20-1711 & 20-1793 released hazardous waste in different quantities and concen- trations at different times over several decades, at least from the 1960s to well into the 1980s. What is more, the record does not establish when or in what amounts any of this pollution occurred. And, as the district court acknowledged in its posttrial findings of fact, once the hazardous chemicals en- tered the groundwater, commingled, and vaporized, the indi- vidual molecules that make up the plume became indistin- guishable. Layer this reality onto the burden a party seeking apportionment faces and the district court’s error becomes clear. Rather than recognizing the rarity of divisible liability in CERCLA cases, the district court instead treated the step- one inquiry as satisfied by a threshold proffer of expert testi- mony by Moran that the harm could be apportioned. The district court’s error also manifests as a closely-related procedural misstep. The error had the effect of all but elimi- nating Moran and Major’s burden at summary judgment un- der the substantive law of CERCLA. Federal Rule of Civil Pro- cedure 56 permits entry of summary judgment only “if the movant shows that there is no genuine dispute as to any ma- terial fact and the movant is entitled to judgment as a matter of law.” Here that was the defendants, Moran and Major, as the parties that sought summary judgment on the question whether the remediation liability was divisible. Recall too that, as a substantive matter under CERCLA, the party con- tending that the court should abandon joint and several lia- bility in favor of divisible liability bears the burden of proving that a reasonable basis for apportionment exists. See Burling- ton Northern, 556 U.S. at 614. The critical takeaway is to recognize that the substantive law governing the parties’ claims (here, § 107(a) of CERCLA) Nos. 20-1711 & 20-1793 23 and the procedural rule controlling whether summary judg- ment is appropriate work in tandem. The Supreme Court un- derscored this precise point in Anderson v. Liberty Lobby, Inc.: “in ruling on a motion for summary judgment, the judge must view the evidence presented through the prism of the sub- stantive evidentiary burden.” 477 U.S. 242, 254 (1986). Stated most succinctly, district courts, in deciding “whether a given factual dispute requires submission to a jury[,] must be guided by the substantive evidentiary standards that apply to the case.” Id. at 255. The district court charted a different course, one that did not account at summary judgment for the substantive burden Major and Moran would face at trial on the apportionment question. The district court even seemed to require Von Duprin to affirmatively disprove that the harm was capable of apportionment. That was error, especially at the summary judgment stage. On the summary judgment record as we see it, there ap- pears to have been a genuine dispute over material facts that bore on the question whether the harm was capable of appor- tionment. The district court should have viewed the trial as the vehicle for resolving—as a definitive matter of evidence— that question. By proceeding otherwise, the district court dis- placed the norm of joint and several liability in favor of ap- portionable harm, a subversion of the substantive law at odds with CERCLA’s central design. The district court’s error is not one we can correct on ap- peal. Having taken our own fresh look at the summary judg- ment record, we see the question whether remediation liabil- ity is capable of apportionment as both contested and diffi- cult. It was hotly contested by Von Duprin and Major and 24 Nos. 20-1711 & 20-1793 Moran—each presented evidence bearing on the questions underlying the apportionment inquiry. And it is difficult be- cause the overarching apportionment question that the par- ties disagree over begets even more questions. The open ques- tions include what role some variables—such as the length of time between the actual discharge of the hazardous chemicals and the chemical analysis of the soil on the properties, the re- moval of soil on the properties as part of past remediation ef- forts, and the commingling of the solvents once they entered the groundwater—might play in deciding the feasibility of determining who caused what harm with any remote confi- dence. Determining whether liability can be apportioned re- quires wrestling with these and other questions presented by the complexity of the underlying facts. Given how fact and context-specific the apportionment in- quiry is—even at step one—we think it wiser to provide these observations and give the district court another opportunity to evaluate the question. In approaching this question anew, the district court should analyze the facts and record evidence at a higher level of particularity under the standards articu- lated in the Restatement (Second) of Torts and the pertinent case law. Do not overread our conclusions. We break no new ground on the law of apportionment in CERCLA cases. We merely restate the high bar for parties arguing that apportion- ment is appropriate in a given CERCLA cost-recovery action. The district court did not take that into account and otherwise failed in its summary judgment ruling to support its decision to depart from the norm of joint and several liability with fac- tual findings rooted in the record. Nos. 20-1711 & 20-1793 25 B. Assignment of Liability After Trial Our reversal of the divisibility determination at summary judgment compels us to vacate the damages award for a fresh liability calculation as a whole. But we need to address an ad- ditional error by the district court. After determining that ap- portionment was appropriate, the district court proceeded to Major and Moran’s request for contribution under § 113(f) and allocated the liability among the parties. While we can see the ultimate liability awards, we cannot tell how the district court weighed the equitable factors. The error, as we see it, resulted from the district court treating apportionment and allocation interchangeably—so much so that we cannot be sure how the court arrived at its determinations of liability. We are not quibbling with word choice. The distinction be- tween apportioned liability and allocated liability is substan- tive and important under CERCLA. These terms are far from synonymous. Indeed, they refer to altogether separate statu- tory provisions. Apportionment operates in a § 107(a) action for the recovery of response costs, but only where the harm is divisible. See Burlington Northern, 556 U.S. at 614. Allocated liability, on the other hand, arises by operation of § 113(f). See Atl. Rsch., 551 U.S. at 140. That provision authorizes PRP de- fendants to seek contribution toward the liability from the party seeking cost recovery in the first instance or from other PRP defendants. As happened here, the defendants in a § 107(a) cost-recovery action typically bring a counterclaim under § 113(f) for contribution from the plaintiff. Allocation refers to the way liability is assigned in a con- tribution claim. See NCR Corp. v. George A. Whiting Paper Co., 768 F.3d 682, 689 (7th Cir. 2014) (“2014 NCR Corp.”). Unlike apportionment, which is guided by principles of causation, 26 Nos. 20-1711 & 20-1793 § 113(f) of CERCLA provides that courts allocate liability among PRPs “using such equitable factors as the court deter- mines are appropriate.” 42 U.S.C. § 9613(f)(1). It seems easiest to think of the allocation analysis as more flexible and softer than the apportionment analysis under § 107(a), which re- quires more an application of principles of causation and a form of contributory negligence. One court has described the distinction more colorfully: “To apportion is to request sepa- rate checks, with each party paying only for its own meal. To allocate is to take an unitemized bill and ask everyone to pay what is fair.” Yankee Gas Servs. Co. v. UGI Utilities, Inc., 852 F. Supp. 2d 229, 241–42 (D. Conn. 2012). The district court exercises substantial discretion when al- locating liability. As Congress put it, district courts must allo- cate response costs using “such equitable factors as the court determines are appropriate,” 42 U.S.C. § 9613(f)(1), a level of discretion we have described as “broad.” 2014 NCR Corp., 768 F.3d at 700. But that discretion is not without limits, and the “court’s ultimate decision must reflect CERCLA’s equitable principles.” Id. at 701. We have previously vacated a district court’s allocation of damages where the district court did not adequately explain its rationale for evaluating some relevant equitable factors and not others. See id. at 702. Just as we ob- served in 2014 NCR Corp., we must vacate the judgment where “we cannot be sure either that the court did, or that it did not, adequately consider all of the circumstances before makings its decision.” Id. at 703. Just so here. After reaffirming its view that the harm was capable of ap- portionment at summary judgment, the district court pro- ceeded to a bench trial to conduct what it called “equitable allocation.” To be sure, the district court mentioned equitable Nos. 20-1711 & 20-1793 27 factors that it considered, including listing the Gore factors, but the court went into no detail as to how or why it consid- ered certain factors or discounted others. Instead, the district court seemed to rely solely on Dr. Love’s expert testimony on how the harm could be apportioned based on the four distinct sources of contaminants. Without an explanation from the district court, we cannot conduct our own evaluation of its ex- ercise of discretion. In many ways, we find ourselves stuck. We cannot discern how the district court determined how much to award each party and on what basis. Even more, though, our review of the district court’s posttrial opinion leaves us of the firm con- viction that the court merged and conflated its consideration of apportionment and allocation. We do not know what por- tions of the liability awards reflect apportionment and what aspects reflect allocation. The Supreme Court cautioned against this error in Atlantic Research, emphasizing that treat- ing “contribution as if it were synonymous with apportion- ment of expenses among PRPs” too “confuses the complimen- tary yet distinct nature of the rights established in §§ 107(a) and 113(f).” 551 U.S. at 138 (internal citation and quotation omitted). The Court sounded similar caution in Burlington Northern, emphasizing that “[e]quitable considerations play no role in the apportionment analysis; rather, apportionment is proper only when the evidence supports the divisibility of the damages jointly caused by the PRPs.” 556 U.S. at 615 n.9. We see no path forward other than to vacate the judgment and allow the district court to revisit the entirety of its liability analysis. On remand and with the legal framework now clar- ified, the district court must first revisit the apportionment question it previously resolved at summary judgment on Von 28 Nos. 20-1711 & 20-1793 Duprin’s § 107(a) cost-recovery claim. Our review of the sum- mary judgment record, as we have explained, leaves us per- suaded that resolving the apportionment inquiry at the very least requires consideration of the evidence presented at trial and perhaps even additional evidence. The district court has the discretion to fashion the remand proceedings in ways that position the evidentiary record to allow for the findings nec- essary to resolve the apportionment question. The court will have the same flexibility in revisiting the allocation of liability when it resolves Major and Moran’s counterclaims under § 113(f). IV Beyond the assignment of liability, the parties raise a host of other issues on appeal. Though none is outcome determi- native, we address them to provide as much clarity as possi- ble for the proceedings on remand. On these remaining is- sues, we see no errors and indeed affirm each of the district court’s challenged rulings. A. Major’s Bona Fide Prospective Purchaser Defense To encourage investment in and restoration of contami- nated lands, Congress provided a complete defense to liabil- ity for parties who would otherwise face liability solely by vir- tue of acquiring and thus owning or operating on contami- nated land. CERCLA calls this the bona fide prospective pur- chaser or BFPP defense. See 42 U.S.C. §§ 9601(40), 9607(q)(1)(C). The district court determined that Major quali- fied for the BFPP defense for two properties (the Moran Prop- erty and the Zimmer Packaging Facility) but not for two oth- ers (the Ertel Property and the Zimmer Paper Facility). Major appeals the district court’s ruling on the latter two properties. Nos. 20-1711 & 20-1793 29 To qualify for the BFPP defense, PRPs must meet certain requirements, including that—in connection with acquiring the property in question—they make “all appropriate inquir- ies into the previous ownership and uses of the facility in ac- cordance with generally accepted good commercial and cus- tomary standards and practices.” 42 U.S.C. § 9601(40). Zimmer Paper Facility. The district court was right to con- clude that Major was not entitled to the BFPP defense for the Zimmer Paper Facility. To be sure, Major did complete a Phase 1 Environmental Assessment according to the provi- sions of ASTM International Standard E1527-05, and, comply- ing with that standard, did at the time satisfy parts of the “all appropriate inquiries” requirements set forth in portions of CERCLA’s implementing regulations. See 40 C.F.R. §§ 312.23- 312.31. But CERCLA’s regulations—in particular 40 C.F.R. §§ 312.21 and 312.22—required more, and that more is where Major fell short. Those provisions include requiring certain attestations about the professional qualifications of the envi- ronmental professionals conducting the inquiry for the pro- spective purchase. See 40 C.F.R. § 312.21(d). These required attestations appear nowhere in Major’s Phase 1 Environmen- tal Assessment, however. Try as Major does to point to non- binding agency guidance, it cannot show full compliance with all requirements in the “all appropriate inquiries” regulation. Ertel Property. The district court was equally right to con- clude that Major did not qualify for the BFPP defense for the Ertel Property. Here, too, Major failed to make all appropriate inquiries, though its missteps related more to timing than substance. Major leased the Ertel Property from the City of Indianapolis in November 2007, purchasing it outright sev- eral years later in 2013. Major completed all appropriate 30 Nos. 20-1711 & 20-1793 inquiries only at the time it acquired title to the property in 2013. We agree with the district court’s conclusion that the BFPP defense, by its terms, applies to “owner[s] and opera- tor[s],” and Major became an operator in November 2007, when it commenced its 99-year lease. See 42 U.S.C. § 9607(a)(1). Major therefore had to complete the required in- quiries by performing or updating an environmental assess- ment within 180 days of becoming an owner or operator. The company failed to do so, for the only environmental assess- ment that could satisfy the BFPP’s substantive requirements comes from September 6, 2006—a date not within 180 days of the commencement of the lease in November 2007. Moran Property. Von Duprin, in its combined reply and cross-appeal brief, adds that the district court erred at sum- mary judgment when it allowed Major to assert the BFPP de- fense for the Moran Property. That position lacks merit. The evidence presented at trial established that Major completed the necessary inquiries in a complete and timely fashion in connection with acquiring the Moran Property. B. Von Duprin’s Compliance with the National Contingency Plan We turn now to Major’s challenges to the district court’s determinations that certain costs incurred by Von Duprin were recoverable under § 107(a) of CERCLA. The parties disagree about the applicable standard of re- view. Major invites us to review the district court’s determi- nation of NCP compliance de novo while Von Duprin urges application of the clear error standard. Our decision in Nu- traSweet suggests (without expressly stating) that our review is only for clear error. See 227 F.3d 776, 791 (7th Cir. 2000) Nos. 20-1711 & 20-1793 31 (explaining that “the district court did not clearly err in con- cluding that NutraSweet had satisfied the NCP”). But we do not need to resolve the question because the district’s conclu- sions satisfy either standard. We see no errors in the district court’s findings that Von Duprin incurred most of the chal- lenged costs in substantial compliance with the NCP. We therefore reject Major’s cross-appeal on this point. Congress permitted the recovery of necessary response costs incurred “consistent with” the National Contingency Plan or NCP, a federal regulation that establishes standards and obligations for remediation and cleanup efforts. 42 U.S.C. § 9607(a)(4)(B); see also NutraSweet, 227 F.3d at 791. “A private party response action will be considered ‘consistent with the NCP’ if the action, when evaluated as a whole, is in substan- tial compliance with the applicable requirements … and re- sults in a CERCLA-quality cleanup.” 40 C.F.R. § 300.700(c)(3)(i). The requirements are numerous and task specific. See, e.g., id. § 300.700(c)(5)–(6). The ultimate and con- trolling inquiry is whether the expenses in question were in- curred in “substantial compliance” with the NCP. Id. § 300.700(c)(3)(i). Perfect compliance is not the measuring stick. When considering whether the costs Von Duprin sought to recover were incurred consistent with the NCP, the district court evaluated the relevant provisions of the regulations for the specific types of costs Von Duprin sought to recover. It concluded that Von Duprin’s remedial efforts, on the whole, reflected substantial compliance with the NCP. This finding, in turn, resulted in the district court determining that Von Duprin incurred the following costs consistent with the NCP: $750,000 for preliminary investigation and site assessment, 32 Nos. 20-1711 & 20-1793 $120,000 for remediation of the public park, $465,000 for resi- dential vapor intrusion remediation, and $365,000 for bench and pilot testing. The only contrary finding came with respect to the $1.5 million settlement Von Duprin paid to Threaded Rod and about $39,000 in IDEM oversight costs. The district court found those costs were not incurred consistent with the NCP. Major contends that the district court erred in concluding that Von Duprin’s $750,000 in investigation and site assess- ment costs were recoverable. As best we can tell, Major seems to take issue with the district court’s finding that these costs relate to site assessment and other investigative work under- taken by Von Duprin. But the position is difficult to follow and not advanced with the clarity and evidentiary support to convince us to reverse. Going further, though, we see no legal infirmity in the district court’s observation that § 107(a) per- mits a company to recover due diligence costs incurred in connection with the investigation of a contaminated site. It stands to reason, the district court sensibly explained, that such initial inquiries are necessary to enable subsequent measures to ensure a CERCLA-quality cleanup, as CERCLA and the NCP both contemplate. See CNH Am., LLC v. Cham- pion Envt’l Servs., 863 F. Supp. 2d 793, 809 (E.D. Wis. 2012) (col- lecting district court cases within our Circuit likewise finding that these preliminary investigative costs are recoverable un- der CERCLA). Nor are we convinced by Major’s insistence that the chal- lenged costs are not recoverable because they are not ex- pressly recognized in the NCP. It would be one thing if Von Duprin incurred costs proscribed by the NCP, as such costs would not be recoverable under the plain terms of § 107(a). Nos. 20-1711 & 20-1793 33 But we see nothing inconsistent—lacking substantial compli- ance—with the NCP to permit Von Duprin to recover due dil- igence site-assessment costs. It seems difficult to call such costs inconsistent with the NCP, to say nothing of CERCLA’s broader objectives. On this record, and especially against Major’s undevel- oped argument on appeal, we cannot conclude that the pre- liminary assessment and investigative costs—those not ex- pressly addressed by the NCP—were not incurred consistent with the NCP. We therefore see no error in the district court’s finding that the $750,000 Von Duprin spent on preliminary investigative measures and site-assessment was recoverable under § 107(a). Major also challenges Von Duprin’s recovery of certain other costs. These costs included the remediation of the public park, residential vapor intrusion remediation, and various tests. The district court took considerable care reviewing the testimony offered by Major and Von Duprin on these costs against the backdrop of the pertinent NCP regulations. While noting that Von Duprin’s compliance was not perfect, the dis- trict court concluded these challenged costs were incurred in substantial compliance with the NCP. We agree. Von Duprin substantially complied with the rel- evant components of the NCP dealing with documentation, health and safety protocols, site investigation, and public in- put for that set of challenged costs. And throughout the effort, the company worked with IDEM. See NutraSweet, 227 F.3d at 791 (citing with favor the cooperation with regulatory agen- cies in discussing compliance with the NCP). To be sure, the district court correctly observed that Von Duprin failed to comply to the letter with certain NCP prescriptions. But to 34 Nos. 20-1711 & 20-1793 reverse the district court on this front would require a level of perfection at odds with Congress’s more limited requirement that recoverable costs only be incurred in “substantial com- pliance” with the NCP. C. District Court’s Admission of Dr. Adam Love’s Expert Testimony We come in closing to Von Duprin’s challenge to the dis- trict court’s denial of its motion to exclude the testimony of Dr. Adam Love, Moran’s expert who offered opinions perti- nent to the apportionment of liability under § 107(a) of CERCLA. “We review de novo whether a district court properly fol- lowed the framework for determining the admissibility of ex- pert testimony.” Schultz v. Akzo Nobel Paints, LLC, 721 F.3d 426, 430–31 (7th Cir. 2013); see also Fed. R. Evid. 702; Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579 (1993). If it did, “we re- view its decision to admit or exclude expert testimony only for an abuse of discretion.” Schultz, 721 F.3d at 431. In Daubert, the Supreme Court explained that Federal Rule of Evidence 702 imposes a gatekeeping responsibility on dis- trict courts to ensure that any proposed expert testimony “is not only relevant, but reliable.” 509 U.S. at 589. We afford dis- trict courts substantial latitude in making the findings neces- sary to fulfill this gatekeeping role. See Jenkins v. Bartlett, 487 F.3d 482, 489 (7th Cir. 2007). Latitude is not a blank check, however, and the court must provide more than just conclu- sory statements of admissibility to show that it adequately performed its gatekeeping function. See Naeem v. McKesson Drug Co., 444 F.3d 593, 608 (7th Cir. 2006). Nos. 20-1711 & 20-1793 35 We start with two points of agreement among all parties. First, Dr. Love offered a relevant opinion. Indeed, he opined on whether and the extent to which each property contributed to the commingled plume—the question at the epicenter of whether liability could be apportioned. To our mind, it would be difficult to identify a question more relevant to the entire litigation. Second, nobody questions Dr. Love’s qualifications, as he plainly has the training and experience necessary to of- fer an opinion pertinent to whether responsibility for the con- tamination can be divided among the PRPs. What the parties tangled over was whether Dr. Love reached his opinions by applying a methodology that inhered with sufficient reliability to satisfy the requirements of Rule 702. He came to his opinions by relying on soil and ground- water data collected during site investigations that occurred primarily from 2005 to 2017. Using this data, he analyzed soil samples at each property and then compared that to the chemical makeup of the groundwater plume. Dr. Love claimed that the distinct chemical compounds in the contam- inated soil he tested at each site provided a way to apportion each property’s contributions to the total environmental harm that is the commingled groundwater plume. At no point did Dr. Love opine on which party was responsible for what amount of the harm. Instead, he offered a view on the propor- tion of the total contamination that was attributable to each property. Von Duprin insists that the district court should have ex- cluded Dr. Love’s opinions because his underlying analysis failed to account for a range of confounding variables that prevented him from making reliable findings. While Von Duprin advances some sound observations about the 36 Nos. 20-1711 & 20-1793 limitations of Dr. Love’s ultimate opinions, we cannot say the district court abused its discretion in admitting the opinions in the first instance. Right to it, Von Duprin’s challenge goes more to the weight (or lack thereof) the district court should have afforded Dr. Love’s ultimate opinions. See Burton v. E.I. du Pont de Nemours and Co., Inc., 994 F.3d 791, 826 (7th Cir. 2021) (“Although Rule 702 places the judge in the role of gate- keeper for expert testimony, the key to the gate is not the ul- timate correctness of the expert’s conclusion but rather the soundness and care with which the expert arrived at her opin- ion.”) (cleaned up). Cross-examination, the presentation of contrary evidence, and an instruction on the burden of proof are the appropriate methods through which Von Duprin could have attacked Dr. Love’s opinions. See Schultz, 721 F.3d at 431. Be careful not to confuse what we are saying here. To con- clude, as we have, that the district court did not abuse its dis- cretion in admitting and considering Dr. Love’s opinions is not to say that those same opinions were sufficient as an evi- dentiary matter to establish the divisibility of cost-recovery li- ability under § 107(a) of CERCLA. Admissibility and eviden- tiary sufficiency are not one and the same. * * * This is a difficult case involving many decades, many par- ties, and many millions of dollars. We see no merit to much of the challenges to the district court’s rulings. But we do agree with Von Duprin that the district court’s summary judgment determination on apportionment requires remand. And we agree with Major that the district court’s ultimate allocation of liability necessitates a remand as well. For these reasons, we AFFIRM in part and VACATE and REMAND in part.